When a private Kura runner node becomes unready, cache dispatches now fall back to the public cache instead of keeping builds routed to the dead endpoint. Previously, a wedged Kura node-port server could cause module-cache downloads to time out because the runner-cache endpoint was a hard override. Now the server tracks a readiness heartbeat for private node-port Kura servers, and once that heartbeat is stale it stops handing out the private endpoint, letting builds fall back to the public cache and keeping CI from stalling.
Hive
Kura runner-cache fallback to public cache when private node is unready
Published
Jun 24, 2026 · 12:42 UTC
Repository
tuist/tuist